Solid-state EPR strategies for the structural characterization of paramagnetic NO adducts of frustrated Lewis pairs (FLPs).

Solid-state EPR strategies for the structural characterization of paramagnetic NO adducts of frustrated Lewis pairs (FLPs) Marcos de Oliveira Jr., Thomas Wiegand, Lisa-Maria Elmer, Muhammad Sajid, Gerald Kehr, Gerhard Erker, Claudio José Magon, and Hellmut Eckert Citation: The Journal of Chemical Physics 142, 124201 (2015); doi: 10.1063/1.4916066 View online: http://dx.doi.org/10.1063/1.4916066 View Table of Contents: http://scitation.aip.org/content/aip/journal/jcp/142/12?ver=pdfcov Published by the AIP Publishing Articles you may be interested in Distance measurements between paramagnetic centers and a planar object by matrix Mims electron nuclear double resonance J. Chem. Phys. 122, 024515 (2005); 10.1063/1.1828435 Rotational spectrum and molecular structure of OCS–N 2 O J. Chem. Phys. 114, 4829 (2001); 10.1063/1.1346637 Rotational spectroscopy and molecular structure of 15 N 2 – 14 N 2 O J. Chem. Phys. 110, 4394 (1999); 10.1063/1.478321 The rotational spectrum and nuclear quadrupole hyperfine structure of CO 2 –N 2 O J. Chem. Phys. 108, 3955 (1998); 10.1063/1.475797 The microwave spectrum and nuclear quadrupole hyperfine structure of HCCH-N 2 O J. Chem. Phys. 107, 2232 (1997); 10.1063/1.474620

This article is copyrighted as indicated in the article. Reuse of AIP content is subject to the terms at: http://scitation.aip.org/termsconditions. Downloaded to IP: 128.171.57.189 On: Thu, 02 Jul 2015 07:56:25

THE JOURNAL OF CHEMICAL PHYSICS 142, 124201 (2015)

Solid-state EPR strategies for the structural characterization of paramagnetic NO adducts of frustrated Lewis pairs (FLPs) Marcos de Oliveira, Jr.,1 Thomas Wiegand,2 Lisa-Maria Elmer,3 Muhammad Sajid,3 Gerald Kehr,3 Gerhard Erker,3 Claudio José Magon,1 and Hellmut Eckert1,4,a) 1

Instituto de Física de São Carlos, Universidade de São Paulo, P.O. Box 369, 13560-970 São Carlos, São Paulo, Brazil 2 Laboratorium für Physikalische Chemie, ETH Zürich, Vladimir-Prelog-Weg 2, 8049 Zürich, Switzerland 3 Organisch-Chemisches Institut, WWU Münster, Corrensstraße 40, D 48149 Münster, Germany 4 Institut für Physikalische Chemie, WWU Münster, Corrensstrasse 30, D 48149 Münster, Germany

(Received 12 February 2015; accepted 10 March 2015; published online 27 March 2015) Anisotropic interactions present in three new nitroxide radicals prepared by N,N addition of NO to various borane-phosphane frustrated Lewis pairs (FLPs) have been characterized by continuous-wave (cw) and pulsed X-band EPR spectroscopies in solid FLP-hydroxylamine matrices at 100 K. Anisotropic g-tensor values and 11B, 14N, and 31P hyperfine coupling tensor components have been extracted from continuous-wave lineshape analyses, electron spin echo envelope modulation (ESEEM), and hyperfine sublevel correlation spectroscopy (HYSCORE) experiments with the help of computer simulation techniques. Suitable fitting constraints are developed on the basis of density functional theory (DFT) calculations. These calculations reveal that different from the situation in standard nitroxide radicals (TEMPO), the g-tensors are non-coincident with any of the nuclear hyperfine interaction tensors. The determination of these interaction parameters turns out to be successful, as the cw- and pulse EPR experiments are highly complementary in informational content. While the continuous-wave lineshape is largely influenced by the anisotropic hyperfine coupling to 14N and 31P, the ESEEM and HYSCORE spectra contain important information about the 11B hyperfine coupling and nuclear electric quadrupolar interaction. The set of cw- and pulsed EPR experiments, with fitting constraints developed by DFT calculations, defines an efficient strategy for the structural analysis of paramagnetic FLP adducts. C 2015 AIP Publishing LLC. [http://dx.doi.org/10.1063/1.4916066]

INTRODUCTION

The development of frustrated Lewis pair (FLP) chemistry has been a milestone in chemical research within the last few years.1,2 In intramolecular FLPs, a Lewis acid center is placed in close spatial proximity to a Lewis base center within the same molecule. Bulky substituents present at both centers reduce the covalent interactions among the Lewis centers (“frustration”), thereby imparting cooperative catalytic activity to the molecule. In particular, FLPs containing borane (Lewis acid) and phosphane (Lewis base) centers exhibit remarkable cooperative binding of small molecules, including the capture of the environmentally problematic nitric oxide.3 The five-membered heterocyclic FLP-NO adducts constitute a family of thermally and chemically stable aminoxyl radicals, which are interesting optical, magnetic, and catalytic materials in their own right. While their liquid-state EPR and solid-state NMR properties have been studied in detail,3–5 no information about the anisotropic electronic hyperfine interaction tensors is available at the present time. In principle, the solid state EPR spectra are expected to be influenced by a multitude of parameters, including (a) the three g-tensor components gx x , g y y , and gz z , (b) three sets of hyperfine interaction tensor components Ax x , A y y , and Az z for the interactions of the

unpaired electrons with the 11B, 14N, and 31P nuclei, (c) three sets of Euler angles relating each of the hyperfine interaction tensors to the g-tensor, (d) two electric field gradient tensor (EFG) principal components each for 11B and 14N, and, finally, (e) two additional sets of Euler angles relating the EFG tensor orientations to the g-tensor. Thus, altogether, the spectrum might be influenced by a total of 31 interaction parameters, rendering any effort of determining them by unconstrained least-squares fitting a ridiculous undertaking. In the present contribution, a feasible strategy is developed and applied for the solid-state EPR characterization of borane/phosphane FLP-NO radicals. This is done on the basis of a detailed comparison of continuous-wave (cw) EPR lineshapes, electron spin echo envelope modulation (ESEEM)6,7 frequencies, and two-dimensional (2D) hyperfine sublevel correlation spectroscopy (HYSCORE)8 in conjunction with calculations from density functional theory (DFT). BASIC THEORETICAL BACKGROUND

In this section, we will provide a brief introduction to the main theoretical concepts of the pulsed EPR methods used in the present work.6,9–12 The static spin Hamiltonian for an electron-nucleus spin pair in the solid state is given by

a)Author to whom correspondence should be addressed. Electronic mail:

[email protected] 0021-9606/2015/142(12)/124201/12/$30.00

142, 124201-1

H = βe H · g · S + g n β n H · I + S · A · I + I · Q · I. (1) © 2015 AIP Publishing LLC


124201-2

de Oliveira, Jr. et al.

J. Chem. Phys. 142, 124201 (2015)

The first two terms are the electronic and nuclear Zeeman interactions, respectively, where g is a symmetric tensor that can be represented by six independent components, for example, its principal values, g x x , g y y , and g z z , and three Euler angles describing the orientation of its principal axes relative to a molecular coordinate system. In this work, the g principal axes frame itself is considered as the molecular frame and all interaction tensors are referred to this frame using the Euler angles. The third term represents the electron-nucleus hyperfine coupling interaction. The hyperfine tensor A has the principal values Ax x , A y y , and Az z . The last term represents the nuclear quadrupole interaction. Q is the nuclear quadrupole tensor and its components in its principal axes system are given by13 Qz z =

να = |νI + A/2| , νβ = |νI − A/2| .

Cq , 2I (2I − 1)

Q z z (1 − η) , 2 Q z z (1 + η) =− , 2

Qxx = − Qyy

(2)

where η = (Q x x − Q y y )/Q z z and Cq = e2qQ n/h are electric field gradient asymmetry parameter and the quadrupolar coupling constant due to the nuclear electric quadrupolar moment Q n interacting with the electric-field gradient at the nuclear site. Although the Hamiltonian (1) can be diagonalized numerically with arbitrary precision for a given set of parameters, a better insight into the origin of the nuclear modulation effect can be initially obtained with a semi-quantitative discussion based on a simpler spin system consisting of one electron spin (S = 1/2) and one nuclear spin (I = 1/2). For this case, the four basis functions |m S , m I ⟩ are written as products of electronic and nuclear parts. In addition, the quadrupole interaction will be entirely neglected, and an isotropic g-matrix and axial hyperfine interaction will be considered. Since EPR studies are normally carried out at high fields, where the amplitudes of the A-matrix components are much smaller than the electronic Zeeman energy, perturbation theory can be employed for the Hamiltonian diagonalization procedure. This allows for a neglect of all the off-diagonal matrix elements which connect different m S states. As a result, the spin Hamiltonian (1) in matrix form consists of two blocks, corresponding to the α (m S = 1/2) and β (m S = −1/2) electronspin states, that can be separately diagonalized giving rise to two sets of eigenvalues. Following this procedure, the nuclear frequencies that correspond to the two electron spin manifolds are given by6  ( να =  νI +   ( νβ =  νI − 

FIG. 1. Pulse sequences used for the ESEEM (a) and HYSCORE (b) experiments.

)2 ( )  1/2 A ∥ 2 2  A⊥ 2 sin θ + νI + cos θ  , 2 2  )2 ( )  1/2 A ∥ 2 2  A⊥ sin2θ + νI − cos θ  , 2 2 

(3)

where A⊥ = Ax x = A y y , A ∥ = Az z , and θ is the angle between the magnetic field and the z principal axis of the A-tensor. For the particular case of an isotropic hyperfine coupling (A⊥ = A ∥ = A), the nuclear transition frequencies reduce to

(4)

The addition of the off-diagonal components of the A-matrix mixes the energy states so that they are no longer pure m I = 1/2 or m I = −1/2 states. This means that the forbidden EPR transitions will have nonzero probabilities, providing the basis of the ESEEM effect. The standard one-dimensional three-pulse ESEEM experiment consists of a stimulated echo, produced by two pulses separated by an evolution time τ and followed by a third pulse after the mixing time T (Figure 1(a)). As T is being incremented, nuclear hyperfine and electron-nuclear magnetic dipole-dipole couplings produce a modulation of the echo envelope. The electron coherence order is zero during this time interval, and the modulation thus arises from nuclear coherences.6 Considering ideal non-selective pulses, the echo modulation formula for the three-pulse ESEEM sequence is given by V3p (τ,T) k 1 − cos 2πυ β τ [1 − cos (2πυα (T + τ))] 4 + [1 − cos (2πυα τ)] 1 − cos 2πυ β (T + τ) , ( )2 1 BνI k= , (5) 2π υα υ β B = A ∥ − A⊥ sin θ cos θ.

=1−

The quantity k is named modulation depth parameter. For the case of an isotropic hyperfine interaction (A⊥ = A ∥ ), or if the magnetic field is oriented along one of the principal axes of the hyperfine tensor (θ = 0 or θ = π/2), the echo modulation disappears (k = 0), since in these cases, the parameter B in the above equation becomes zero. An important feature of the three pulse ESEEM is the dependence of the modulation amplitude on τ, as is clear from the factors 1 − cos(2πυ β τ) and 1 − cos(2πυα τ). For να, β = n/τ (n = 1, 2, . . .), the modulation at frequency νβ,α vanishes. Thus, the three pulse ESEEM spectra are subject to “blind spots,” which become denser for increasing evolution time τ. Consequently, the three pulse ESEEM experiment has to be performed at more than one τ value, in order to ensure that such blind spots are avoided. The Fourier transformation of the modulation yields the ESEEM spectra. In the limit of low modulation depth, these spectra can be approximately described by a superposition of


124201-3


all the να, β components associated with the different nuclei to which the electron spins are coupled.6 In the present work, we investigate in detail the properties of the ESEEM spectra for the case I = 3/2, which represents a more complex situation than the one described above. In general, the analytical diagonalization of the Hamiltonian for I > 1/2 is considerably more difficult and, usually, the ESEEM spectra must be calculated by numerical diagonalization of the entire Hamiltonian matrix, avoiding perturbation procedures. In randomly oriented solid samples, the ESEEM spectrum is a superposition of sub-spectra from many different orientations each with different nuclear frequency and different modulation depth. In general, due to the strong angular dependence of the state mixing, and thus the modulation depth, a spectrum from which the principal values of the hyperfine coupling tensors can be easily assigned, cannot be obtained. Therefore, computer simulations are required for the proper analysis of the spectra. Reijerse and Keijzers14 have investigated the dependence of ESEEM powder line shapes for the cases with electron spin S = 1/2 and nuclear spins I = 1/2, 1, and 3/2 on the various coupling parameters. In the case of I = 3/2, they have performed simulations for a generic system with νI = 1 MHz, Cq = 0.4 MHz, and variable hyperfine coupling constant. Yiannis et al.15 investigated the time and frequency domain ESEEM spectra of paramagnetic centers in a γ-irradiated B2O3 glass. They carried out a set of simulations considering electron-nuclear hyperfine interactions with the 11 B nucleus. These simulations were done for the limiting case in which the hyperfine coupling constant Aiso = (Ax x + A y y + Az z )/3 is smaller than twice the nuclear Zeeman frequency, 2νI . For the present FLP systems studied here, the situation is rather different: from liquid-state cw-EPR3 and DFT calculations Aiso is known to be around −10 MHz, whereas the 11 B nuclear Larmor frequency under X-band EPR conditions is νI = 4.57 MHz. Thus, the restriction, |νI − | Aiso/2|| < 0.1 × νI applies for this case.14 As this particular situation has not been analyzed for spin-3/2 nuclei to the present date, we have carried out systematic simulations to investigate the influence of the quadrupolar and hyperfine interaction parameters on the ESEEM spectra. The HYSCORE technique is a powerful tool for the investigation of weak hyperfine and nuclear electric quadrupolar couplings of paramagnetic centers in disordered systems. This experiment is quite well understood for systems in which an unpaired electron (S = 1/2) interacts with I = 1/216–18 or I = 1 nuclei.19,20 In contrast, there is little information in the literature regarding HYSCORE involving spin I > 1 systems. Of great interest is the work of Pöppl et al.21 who have studied the spectral features of the HYSCORE spectrum of disordered S = 1/2, I = 3/2 systems. They have analyzed the influence of the hyperfine and quadrupolar interactions on the cross peak line shapes due to single- and double-quantum transitions. Particularly, they have demonstrated that the lines representing the double-quantum transitions are of special use for the determination of the principal values of the hyperfine and quadrupolar coupling interactions.21 The present contribution expands upon this work, including simulations for more strongly coupled systems, allowing the nuclear quadrupolar interaction parameters to be deduced from such data.

J. Chem. Phys. 142, 124201 (2015)

In the two-dimensional HYSCORE experiment, the nuclear spin transitions in the two different m S spin manifolds are correlated to each other by non-diagonal cross-peaks, appearing at (να , νβ ), (νβ , να ) and (−να , νβ ), (−νβ , να ), respectively, in the (+, +) and (−, +) quadrants of the 2Dspectrum.6,17 In the limit of a weak hyperfine interaction, the contributions with positive phase modulation dominate and the cross-peaks appear predominantly in the (+, +) quadrant.6,15,22 In the opposite limit of a strong hyperfine interaction, the contributions with negative phase modulation dominate and the cross-peaks appear predominantly in the (−, +) quadrant. Near the cancellation condition (Aiso ∼ 2νI ), the cross-peaks in the 2D-HYSCORE spectrum have comparable intensities in both quadrants.6,15 EXPERIMENTAL SECTION Sample preparation and characterization

The compounds 1, 2, and 3 were obtained analogously as previously described by us for the preparation of the FLPNO radicals 1 or 3 and their corresponding FLP-NOH derivatives (Scheme 1).2–4 In a first step, the pure nitroxide radical compound is formed via N,N addition of nitric oxide by the parent frustrated Lewis pair compound in fluorobenzene solution at room temperature. The resulting aminoxyl radical undergoes rapid H-abstraction reactions with 1,4 cyclohexadiene in benzene solution, resulting in the diamagnetic hydroxylamine compounds. As this reduction is never 100% complete, these samples always contain a small amount of the original paramagnetic radical species, which presumably substitute on the molecular sites within the crystal structure of the host compound. As the EPR spectra of pure solid radical samples are excessively broadened due to strong intermolecular magnetic dipole-dipole interactions, the measurements were done on the magnetically dilute samples of the corresponding diamagnetic polycrystalline hydroxylamine hydrogenation products. The crystal structures of the FLP-NOH host compounds of 1 and 3 have already been published,3(a),3(b),4 whereas the crystallographic characterization of compound 2 will be subject of a forthcoming publication. The solid-state NMR characterization of 1 and 3 and their hydrogenation products has also been published.3(b),5

SCHEME 1. Generation of the dilute FLP-NO radicals in the respective diamagnetic FLP-NOH matrix.


124201-4


Solid-state EPR

Continuous-wave length and pulsed solid-state EPR spectra were measured at 100 K on a E-580 BRUKER ELEXIS X-band EPR spectrometer. For the acquisition of the cw spectra the following parameters were used: Microwave power ∼1 mW, sweep rate 3.6 G/s, time constant 40.96 ms, 1024 points modulation frequency and amplitude 100 kHz and 2 G, respectively. ESEEM spectra were obtained at 100 K under an external field strength of 3350 G using the three-pulse sequence (t p ) − τ − (t p ) − T − (t p )—echo shown in Figure 1(a), with a π/2 pulse t p = 6 ns. The delay between first and second pulse, τ, was set to 170 ns or 250 ns (the possible occurrence of blind spots was discarded after measurements using different τ values). The time interval T was incremented in 16-ns steps starting with T = 300 ns. 100 acquisitions were accumulated for each increment with repetition times of 300 µs and up to 20 scans were added up for signal averaging. The single echo response was measured with the commonly used four step phase cycling of the first and second pulse to avoid unwanted primary echoes and FID distortions.6 The resulting data were processed in the following way: the modulated echo decay was fitted to a biexponential function which in turn is subtracted from the experimental result in order to remove the pure decaying component. Following further apodization and zero-filling, the oscillating signal was Fourier-transformed, resulting in the ESEEM spectrum. The echo detected absorption spectra (available as supplementary material42) were recorded using the three-pulse sequence. The integrated echo intensities were measured as a function of the magnetic field strength over a range of 3270–3430 G. The pulse spacing between the first two pulses (t) was set to 250 ns, and the time between the second and the third pulse was 4000 ns in order to suppress nuclear modulation effects. HYSCORE experiments were conducted at an external magnetic field strength of 3350 G using the pulse sequence (t p ) − τ − (t p ) − t 1 − (2t p ) − t 2 − (t p )—echo shown in Figure 1(b), with τ = 250 ns. The echo intensity was measured as a function of t 1 and t 2, where t 1 and t 2 were incremented in steps of 24 ns from the initial values of 300 ns. Pulses of 6 ns for the π/2 pulse and 12 ns for the π pulse were used to record a 128 × 128 matrix. The application of a 4-step phase cycling procedure was used to eliminate the unwanted echoes. Solid state powder EPR data were simulated by the functions “pepper” (cw) and “saffron” (ESEEM and HYSCORE) of the software package EasySpin23 implemented in MATLAB (MathWorks, Inc.). Multi-component least-squares fitting was performed by means of the function “esfit” belonging to the same package. DFT calculations

The geometry of a single molecule of compound 1 was taken from a fully geometry optimized structure obtained by Grimme and coworkers,4 whereas in case of compounds 2 and 3 full geometry optimizations in the gas phase on a DFT metaGGA (TPSS24) level of theory applying the D3(BJ) dispersion correction25 and Ahlrich’s def2-TZVP26 basis set were performed. In additional calculations, the geometries of 2 and 3

J. Chem. Phys. 142, 124201 (2015)

were taken from the single crystal structures and only the H atoms were optimized on the same level of theory as mentioned before. All geometry optimizations were performed within the TURBOMOLE program suite.27 In all TURBOMOLE SCF calculations, an energy convergence criterion of 10−7 Eh and in all geometry optimizations an energy convergence criterion of 5 × 10−7 Eh was chosen. The integration grid was set to m428 and the RI approximation29 was used. EPR parameters such as hyperfine tensors and g-tensors were calculated with the ADF software package.30 11B, 14N, and 31P hyperfine tensors were calculated with the B3LYP31 DFT functional using TZ2P and QZ4P32 basis sets (all electron basis sets) and visualized with the software DIAMOND.33 The INTEGRATION keyword was set to 6.0 and in the SCF calculation an energy convergence criterion of 10−6 Eh was used. The g-tensors were determined on a B3LYP/TZP level of theory using spin-orbit coupled spin unrestricted relativistic ZORA calculations.34 11 B and 14N nuclear electric quadrupolar coupling parameters were calculated on a B97-D/def2-TZVP level of theory with the GAUSSIAN program suite (version GAUSSIAN09).35 In case of 11B a slightly modified version of the def2-TZVP basis set with tighter basis functions for the boron atoms was used (for more details see Ref. 36). The corresponding output files were analyzed with the EFGShield tool.37

RESULTS AND DISCUSSION Continuous-wave EPR

Figure 2 shows the X-band cw-EPR spectra for the set of samples 1-3 examined (black curves) and the best simulations obtained for each spectrum (red curves). The signals shown are the only ones observed within the entire field range (0 to 6 kG). Although the spectra closely resemble each other, small differences can be observed. Tables I and II summarize the fitting parameters obtained by a procedure described in more detail below. Our results indicate that the details of the FLP backbone have only a small, even though non-negligible effect upon the spectra. This view is also supported by the DFT calculations done on the samples examined. Simulation of these cw-EPR spectra constitutes a considerable many-parameter challenge. First of all, test simulations conducted on our materials and also on the reference material TEMPO indicate that the 14N and 11B nuclear electric quadrupolar interactions with coupling constants predicted by our DFT calculations have no appreciable influence on the cwEPR lineshapes (see sections in supplementary material42). In addition, the simulations show that the 11B nuclear hyperfine interaction (with a value of Aiso near −10 MHz as calculated by DFT on a B3LYP/TZ2P level of theory) has only a very small effect on the EPR lineshape. Thus, the cw-spectrum is dominated by the g-tensor, as well as the 14N and 31P hyperfine coupling tensors and their relative orientations. Based on these preliminary calculations, three approaches were taken for the simulations as summarized in Figure 2. In parts (a), (d), and (h) of this figure, the DFT calculated values (from gas phase optimization) are used, but the tensors are assumed to be coincident. In parts (b), (e), and (i), these DFT-calculated values are used


124201-5


J. Chem. Phys. 142, 124201 (2015)

FIG. 2. Comparison between simulated (red lines) and experimental spectra (black lines) for samples 1 ((a)–(c)), 2 ((d)–(g)), and 3 ((h)–(k)). Simulations do not include the quadrupole couplings. Simulations ((a), (d), (h)) use the DFT parameters calculated for the gas phase, with diagonal A-tensors; ((b), (e), (i)) same as ((a), (d), (h)), but including a relative A- and g-tensor orientation; ((f), (j)) simulations consider the DFT parameters calculated from crystallographic data for samples 2 and 3; ((c), (g), (k)) simulations are obtained after optimization of the principal values for A- and g-tensors using the DFT results of ((b), (e), (i)) as starting parameters.

as well, and the calculated relative orientations between the Aand g-tensor principal axes are included in the simulations. This information can be expressed through the relative Euler angles corresponding to the transformation from the A- to the g-tensor principal axes systems. The angles were obtained following the definition in Ref. 6 and their values are shown in Table II. Table II illustrates that the angle β defining the orientation of Az z relative to gz z is almost zero for 14N, near 20◦ for 31P, and near 90◦ for 11B in all three FLPs studied. Somewhat larger variations are observed for the angles α and γ. Scheme 2 shows the representation of the principal axis systems for compound 1, following the usual convention for nitroxide radicals, with gz z perpendicular to the NO bonding direction (this convention is different from that used for the ADF calculations, where gz z and −gx x directions are interchanged). The comparison of Figures 2(a), 2(d), and 2(h) with Figures 2(b), 2(e), and 2(i) indicates that the relative orientations between the A- and g-tensors have an appreciable effect on the cw-EPR spectral lineshapes and need to be considered when simulating the experimental data. Also included (Figures 2(f) and 2(j)) are comparisons of the experimental spectra with simulated data based on DFT-calculations from crystallographic input data for compounds 2 and 3. While in the case of 3, the results match closely to those from the gas phase calculation, a considerable discrepancy between both of these DFT data sets was noted for compound 2, producing a strong deviation between the simulated and the experimental EPR spectra if crystal structure

input parameters are used. Based on this disagreement, the crystal structure based DFT calculation for compound 2 was discounted, and best-fit simulations were obtained for all three compounds using the DFT values from the gas-phase calculations as starting parameters, followed by a manual optimization process for agreement between simulated and experimental lineshapes. In this process, the g- and A-tensor component values were varied systematically in small steps, keeping the giso and Aiso values as close as possible to the experimental values observed in the liquid state cw-EPR spectra previously recorded3 and the DFT-calculated values (Table III). The final results are shown in Figures 2(c), 2(g), and 2(k). Note that these simulations also include the 11B-hyperfine tensor information probed in the pulsed EPR experiments described below. Table I summarizes the A- and g-tensors used for the simulation of the cw-EPR spectra, based on the DFT-calculated Euler angles of Table II. Note that the g components and 14N hyperfine interaction parameters differ substantially from those published for common nitroxide radicals,38–40 indicating that binding to the borane and phosphane moieties profoundly alters the electronic structure of the NO moiety. ESEEM spectra

Complementary information comes from the pulsed EPR (ESEEM) spectra recorded at 3350 G. Figure 3 shows the results for compounds 2 and 3 at 100 K for two different τ

TABLE I. EPR interaction parameters obtained from DFT calculations (geometry-optimized structures from the gas phase on a TPSS-D3/def2-TZVP level of theory, the A tensors were calculated on a B3LYP/TZ2P, the g-tensors on a B3LYP/TZP level of theory) and from the simulation of the cw-EPR spectra in the FLP-NO radicals 1, 2, and 3. 14N

31P

(MHz)

11B

(MHz)

(MHz)

Sample/method

Ax x

Ay y

Az z

Ax x

Ay y

Az z

Ax x

Ay y

Az z

gx x

gy y

gz z

TEMPO/cw (1)/DFT (1)/cw (2)/DFT (2)/cw (3)/DFT (3)/cw

20 −4.0 −4.1 −3.4 −0.5 −3.0 −4.7

20 −3.3 −3.6 −2.8 −1.1 −2.5 −5.3

104 53.5 52.2 56.4 58.7 58.1 61.0

... −54.7 −57 −56.5 −57 −56.5 −61

... −48.3 −51 −48.9 −51 −48.4 −41

... −41.7 −45 −43.3 −43 −43.2 −45

... −11.4 −11 −11.4 −11 −11.0 −11

... −11.4 −11 −11.2 −11 −10.9 −11

... −6.6 −6.7 −6.5 −6.5 −6.2 −6.3

2.0084 2.0161 2.0150 2.0153 2.0161 2.0149 2.0138

2.0053 2.0072 2.0052 2.0071 2.0055 2.0072 2.0065

2.0019 2.0019 2.0019 2.0019 2.0018 2.0019 2.0018


124201-6


J. Chem. Phys. 142, 124201 (2015)

TABLE II. Euler angles corresponding to the A-tensor principal axis system orientation relative to the g-tensor calculated from DFT. Sample

Nuclei

α (deg)

β (deg)

γ (deg)

14N

−19.1 −32.6 62.1

1.2 26.8 89.2

−86.0 64.5 −6.4

−73.7 6.3 −55.1

9.9 16.9 95.8

7.0 17.9 65.1

−30.5 47.6 −65.0

0.5 21.4 88.1

−79.1 −13.0 62.1

−26.7 −49.9 −55.8

0.6 25.9 90.8

34.5 5.6 61.6

−41.8 −8.9 −61.9

0.3 21.0 91.1

0 45.8 64.0

31P

(1)a

11B 14N 31P

(2)b

11B 14N 31P

(2)a

11B 14N 31P

(3)b

11B 14N 31P

(3)a

11B a Based b Based

on gas phase optimization. on single crystal data.

values (170 and 250 ns); no data could be recorded for compound 1 owing to very short spin-spin relaxation times. Peaks corresponding to distant 19F and 1H nuclei are located at 13.46 and 14.25 MHz. ESEEM simulations based on the 14N, 31P, and 11B hyperfine parameters of Table I were found to be identical to those based on the hyperfine parameters of 11B alone, indicating that the ESEEM spectra are completely dominated by the 11B, 19F, and 1H hyperfine and quadrupole couplings. Evidently, the echo envelope modulation caused by the strong anisotropic hyperfine coupling with 14N and 31P produces very strong inhomogeneous broadening in the frequency domain, making the signal undetectable. Also, the cross-suppression effect,41 arising as a consequence of the product rule6 may contribute to this disappearance, if the modulations of 14N and 31 P turn out to be significantly more shallow than those of 11B. Overall, the ESEEM spectra contain information selectively about the spin Hamiltonians of the more weakly coupled nuclei 11 B, 19F, and 1H, which do not influence the solid-state cwEPR spectra appreciably. Figure 3 reveals a variety of charac-

teristic features. The nuclear Zeeman frequencies of 1H and 19F indicate through-space interactions with protons and fluorine atoms in the weak coupling limit. The protons come from the FLP backbone and the residues attached to the P atoms, while the 19F nuclei belong to the substituents on the boron atoms. As all of them are relatively distant from the unpaired electron and their hyperfine coupling constants are thus negligible, the ESEEM effect must be attributed to through-space dipolar interactions. Such interactions have been previously probed by solidstate NMR experiments on compound 3 in which some 1H and 19 F resonances show significant pseudo-contact shifts.5 Using systematic simulations shown below, the remaining features near 1.1, 2.2, 3.5, and 10 MHz can all be attributed to electronnuclear hyperfine interactions with the 11B isotope. For this I = 3/2 nucleus, there are a total of six nuclear Zeeman transitions for each electron spin manifold, comprising three | ∆m I | = 1, two | ∆m I | = 2, and one | ∆m I | = 3 transitions. Before we attempt to simulate the experimental spectra, we explore the effects of the various parameters of the spin Hamiltonian for I = 3/2 in order to identify those parameters which determine the main characteristics in the ESEEM spectra. Although the relative abundance of 10B isotope is not negligible (19.8%), its modulations are not observed in the ESEEM spectra of these FLP samples (Again, ESEEM simulations done for an 11 B isotopically pure sample are essentially identical to simulations done for the natural isotopic composition). This fact can be explained by the relatively strong 10B quadrupolar coupling (Cq ∼ 3 MHz), significantly exceeding the 10B nuclear Zeeman frequency (νI = 1.53 MHz) at the field strength investigated. This renders the anisotropic 10B quadrupolar interaction dominant in the spin Hamiltonian of the corresponding molecule. Owing to its angular dependence in polycrystalline samples, the 10B ESEEM signal is then broadened beyond detectability in the presence of the 11B based signals.15 Figures 4(a) and 4(b) show simulated ESEEM spectra considering an isotropic hyperfine coupling with no quadrupolar interaction for 11B. Two τ values are considered in the pulse sequence to avoid possible blind-spots in the simulations (τ = 170 and 250 ns, respectively). As transitions with | ∆m I | > 1 are not allowed for the isotropic case, all the spectra exhibit just two resonance lines at the frequencies να and νβ given by Eq. (4).6 The corresponding HYSCORE spectra for this simplified system are shown in Figures 4(c), 4(d) and 4(e), simulated for Aiso values of −1.7, −9.7, and −10.7 MHz, respectively. When | Aiso/2| < νI , there are cross peaks at positions (να , νβ )

SCHEME 2. Relative orientations between the principal axis systems of the g-tensor and the A-tensor for 14N (a), 31P (b), and 11B (c) in compound 1.


124201-7


J. Chem. Phys. 142, 124201 (2015)

TABLE III. Isotropic A and g values for the investigated compounds obtained by DFT calculations (geometry-optimized structures from the gas phase on a TPSS-D3/def2-TZVP level of theory, the A tensors were calculated on a B3LYP/TZ2P and the g tensors on a B3LYP/TZP level of theory), and from experimental simulation of solid state cw-EPR spectra. Values obtained from Ref. 3 for cw-EPR in the liquid state (fluorobenzene solution, 300 K) are also shown. A values are in MHz. A iso(14N)

Aiso(31P)

A iso(11B)

giso

Sample/method

DFT

Solid

Liquid

DFT

Solid

Liquid

DFT

Solid

Liquid

DFT

Solid

Liquid

(1) (2) (3)

15.4 16.8 17.5

15 17 17

19 ... 21

−48.2 −49.5 −49.3

−51 −48 −49

49 ... 51

−9.8 −9.7 −9.4

−9.6 −9.5 −9.4

9.1 ... 9.0

2.0084 2.0081 2.0080

2.0074 2.0078 2.0074

2.0089 ... 2.0083

and (νβ , να ) in the (+, +) quadrant of the HYSCORE spectra (Figure 4(c)). In contrast, for the case | Aiso/2| > νI the cross peaks appear at the positions (−να , νβ ) and (−νβ , να ), i.e., in the (−, +) quadrant of the HYSCORE spectra (Figure 4(e)). In the case of | Aiso/2| ∼ νI the cross peaks split in the two quadrants of the HYSCORE spectra (Figure 4(d)). In such simple cases the isotropic hyperfine coupling constant can be experimentally determined from the peak positions. We describe the anisotropy of the A-tensors in terms of the parameters Ax x + A y y , δ = Az z − 2 (6) Ax x − A y y ηA = . Az z − Aiso For axial symmetry (Ax x = A y y ), η A is equal to zero. Figure 5 shows simulated ESEEM spectra as a function of the 11B hyperfine coupling anisotropy parameter δ, considering a fixed value of Aiso, obtained from DFT calculations (B3LYP/TZ2P) for sample 2 (Aiso = −9.7 MHz) and a value of zero for the asymmetry parameter η A, which is in close agreement with the DFT calculation (η A = 0.03). For δ = 0, we have again the isotropic case and the spectrum is composed of two main peaks at 0.28 MHz (να ) and 9.45 MHz (νβ ), according to Eq. (4). When δ , 0 and the A-tensor has axial symmetry, the nuclear transition frequencies in case of spin I = 1/2 are given by Eq. (3). Considering I = 3/2, for negligible quadrupolar coupling interaction, all | ∆m I | = 1 transitions make comparable contributions to the ESEEM spectra, and the resonance frequencies can be approximated by the I = 1/2 case, as discussed below. Figure 6 shows the comparison between the single quantum frequencies obtained from the simulated spectra in Figure 5 and the frequencies obtained from direct calculation using Eq. (3). One can see that, in the absence of quadrupolar

coupling, the approximation for the single quantum transitions to spin I = 1/2 transitions is valid even for highly anisotropic A-tensors. With increasing δ, however, the | ∆m I | > 1 transitions become allowed due to state mixing.6,15,22 For δ = 1.5 MHz, the | ∆m I | = 2 transitions are resolved in the α-spin manifold (small peak around 1 MHz), and for δ ≥ 3.0 MHz all transitions are resolved. On the other hand, only a powder pattern corresponding to the | ∆m I | = 1 transitions is observed for the β-spin manifold, enabling the determination of the hyperfine coupling components by simulation. The best agreement between experimental and simulated spectra is obtained for δ = 4.5 MHz, corresponding to A⊥ = −11.2 MHz and A ∥ = −6.7 MHz (considering Aiso = −9.7 MHz and η = 0). Table III summarizes all the isotropic tensor parameters extracted from the anisotropic simulations and compares them with those measured previously from liquid-state EPR3 and from DFT calculation, documenting an overall consistent picture emerging from all these measurements. As discussed further below, the ESEEM results are also consistent with the results from corresponding HYSCORE studies. For a complete theoretical description, the 11B quadrupolar coupling interaction must be included in the ESEEM simulations. Considering the above-deduced best-fit parameters for the A-tensor, simulations were performed for two τ values, varying the quadrupolar coupling constant (Cq ) for the axially symmetric case (η = 0), assuming both tensors to be coincident in the molecular axis frame. The results are shown in Figure 7. For both τ values, there are resonance lines around 1 MHz and 10 MHz whose positions are not affected by the magnitude of the nuclear electric quadrupolar coupling constant Cq , and according to the observed resonance frequencies these lines can be attributed to the central (m I = −1/2 ↔ m I = +1/2) transitions of the α- and β-spin manifolds, respectively. Evidently, these central transitions are still good for the experimental estimation of the hyperfine coupling components, even when

FIG. 3. Experimental three-pulse ESEEM spectra of samples 2 ((a), (c)) and 3 ((b), (d)) (see Scheme 1) measured at 100 K for ((a), (b)) τ = 170 ns and ((c), (d)) τ = 250 ns.


124201-8


J. Chem. Phys. 142, 124201 (2015)

FIG. 4. ((a), (b)) Simulated ESEEM spectra considering isotropic hyperfine interaction between an unpaired electron and 11B nucleus, for various A iso values and for two different τ values (a) 170 ns and (b) 250 ns. Dashed lines represent the 11B nuclear Zeeman frequency (νI = 4.58 MHz). The spectra are normalized by the maximum intensity. ((c)–(e)) Simulated HYSCORE spectra considering an isotropic hyperfine interaction between an unpaired electron and the 11B nucleus, for various Aiso values: (c) A iso = −1.7 MHz, (d) Aiso = −9.7 MHz, and (e) A iso = −10.7 MHz and τ = 250 ns. Dashed lines represent the ν 1 = ν 2 axes.

the quadrupolar interaction is considerable. The other ESEEM signals show rather complex behavior as a function of Cq and it is difficult to make assignments for all the spectral lines. However, the | ∆m I | = 3 transition line in the α-spin manifold (peak around 3.5–3.9 ppm) has little overlap with other lines and its intensity and central frequency are rather sensitive to variations in the Cq value, making it suitable for estimating this parameter. Another interesting feature that shows up for increasingly strong quadrupolar coupling is a phase inverted peak around 8.5 MHz, which is more evident in the simulations for the spectra recorded at τ = 250 ns than at τ = 170 ns (Figure 7). Up to Cq = 1.5 MHz the intensity and FWHM of this peak increases with increasing Cq . Comparing experimental spectra and simulations, we estimate an imprecision in the determination of Cq of around 0.1 MHz. The simulated spectra that best fit the experimental data for both τ values are the ones with Cq = 1.2 and 1.3 MHz. The last value agrees with the value obtained by 11B MAS-NMR5,36 on the diamagnetic host compound and also with DFT calculations. Using this Cq value, ESEEM spectra were simulated varying η, and the result is shown in Figure 8. The intensity of the peak around 3.5 MHz

(| ∆m I | = 3 transition) is sensitive to variations in η value, while all the other peaks are virtually unaffected. Figure 9 shows simulated ESEEM spectra varying the Euler angle β of the Q-tensor principal axis systems relative to the A-tensor. Clear line shape changes are evident. As β approaches 90◦, the ESEEM spectra get more complicated, with many different lines in the α-spin manifold. Even so, the triple quantum transition around 3.5 MHz shows only an increase in the relative intensity for increasing β. From DFT calculations, we obtain β = 7◦ in sample 2, and the simulations show only small spectral differences over the range β = 0◦ to β = 10◦. Therefore, the approximation of parallel tensors made in the former simulations remains valid for interpretation purposes. The α and γ Euler angles also have a considerable effect on the ESEEM spectrum lineshape when β , 0◦. Although systematic simulations as a function of these angles are not shown here they were carried out for the optimization of the final results shown in Figure 10. Parts (a1), (b1), (c1), and (d1) of Figure 10 show the comparison between ESEEM experimental data and the simulations obtained taking the 11B parameters calculated from


124201-9


FIG. 5. Simulated ESEEM spectra considering an axial hyperfine interaction tensor between an unpaired electron and 11B nucleus. Aiso is fixed at −9.7 MHz and the anisotropy parameter δ (defined in Eq. (6)) is varied between 0 and 12 MHz. The simulations were performed for two different τ values (a) 170 ns and (b) 250 ns (see Figure 1 for the definition of τ). Dashed lines represent the frequencies observed for the main peaks in the experimental spectra. The spectra are normalized by the maximum intensity.

DFT (see Table IV) on a B3LYP/TZ2P level of theory. The comparison is done for two different τ values. For the sample 2, the DFT calculations are in good agreement with the experimental data, as shown in Figures 10(a1) and 10(c1). On

FIG. 6. 11B single quantum nuclear transition frequencies ν β (a) and ν α (b) as a function of the hyperfine anisotropy δ extracted from ESEEM simulated spectra in Figure 5 (symbols), and calculated ESEEM frequencies from Eq. (3) (lines), for two different relative orientations between the applied magnetic field and the z principal axis of the hyperfine coupling tensor: parallel (crosses and black lines) and perpendicular (empty circles and dashed lines).

J. Chem. Phys. 142, 124201 (2015)

FIG. 7. Simulated ESEEM spectra considering an axially symmetric quadrupolar coupling interaction for 11B with variable C q and an axial hyperfine interaction tensor between an unpaired electron and 11B nucleus, with fixed A iso and δ parameters at −9.7 MHz and 4.5 MHz, respectively. Both tensors are assumed to be coincident. The simulations were performed for two different τ values (a) 170 ns and (b) 250 ns (see Figure 1 for τ definition). Dashed lines represent the frequencies observed in the experimental spectra. The spectra are normalized by the maximum intensity.

the other hand, small deviations are observed for the sample 3 (Figures 10(b1) and 10(d1)). Parts (b2) and (d2) show the simulations based on the DFT parameters from the crystal structure of compound 3; no corresponding comparison was done for compound 2 because of the large deviations that were already observed in the cw-EPR spectra. Finally, parts (a2), (b3), (c2), and (d3) of Figure 10 show ESEEM simulations that best fit the experimental data (also including the 1H and 19F Zeeman frequencies). The resulting best-fit parameters are in good agreement with the DFT calculations. These parameters, including all the orientation information are summarized in Table IV.

FIG. 8. Simulated ESEEM spectra assuming Cq (11B) = 1.3 MHz and variable η, and an axially symmetric 11B hyperfine interaction tensor, with fixed A iso and δ parameters of −9.7 MHz and 4.5 MHz, respectively. The simulations were performed for two different τ values (a) 170 ns and (b) 250 ns (see Figure 1 for the definition of τ). Dashed lines represent the frequencies observed experimentally. The spectra are normalized by the maximum intensity.


124201-10


J. Chem. Phys. 142, 124201 (2015)

FIG. 9. Simulated ESEEM spectra considering an anisotropic quadrupolar coupling interaction for 11B, with C q = 1.3 MHz and η = 0.57, and an axial electron-nucleus hyperfine interaction tensor for the 11B nucleus, with A iso = −9.7 MHz and δ = 4.5 MHz. The simulations were performed as a function of the β Euler angle relating the principal axis systems of the A and Q tensors, for two different τ values (a) 170 ns and (b) 250 ns (see Figure 1 for τ definition). Dashed lines represent the frequencies observed for the main peaks in the experimental spectra. The spectra are normalized by the maximum intensity.

Considering the best-fit parameters, obtained from the ESEEM simulations, HYSCORE spectra were simulated, as shown in Figure 11. Figure 11(a) considers only the hyperfine coupling interactions for 11B (Aiso = −9.7 MHz and δ = 4.5 MHz), 1H (Aiso = 1.8 MHz and δ = 0), and 19F (Aiso = 0.8 MHz and δ = 0) and no quadrupolar coupling. In Figure 11(b), the simulation is based on the same hyperfine coupling parameters, including the 11B quadrupolar interaction (Cq = 1.3 MHz and η = 0.57) as well. This simulation is in better agreement with the experimental HYSCORE spectra shown in Figures 11(c) and 11(d). The signals at low frequency are spectral artifacts due to imperfections in the subtraction of the echo decay component in the time domain. The cross peaks around ν1 = ν2 = 6.0, 6.8, 8.2, and 10 MHz are well known artifacts due to pulse imperfections.23 There is also a cross peak

FIG. 10. Experimental (black curves) and simulated (red curves) ESEEM spectra obtained with τ = 170 ns ((a), (b)) and τ = 250 ns ((c), (d)) for the samples 2 ((a), (c)) and 3 ((b), (d)). Simulations (a1), (b1), (c1), and (d1) consider EPR parameters from DFT calculations based on the optimized gas phase structures; simulations (b2) and (d2) consider EPR parameters from DFT calculations based on crystallographic information for compound 2; and simulations (a2), (b3), (c2), and (d3) are the best fits for the experimental data. The simulations where performed considering isotropic hyperfine coupling tensors for 1H and 19F nuclei with Aiso = 1.8 MHz and 0.8 MHz, respectively, and hyperfine and quadrupolar coupling for 11B using parameters shown in Table IV.

around 3.5 MHz that can be attributed to hyperfine coupling interaction with 13C nuclei in the weak coupling limit. Figures 10 and 11 show that both the ESEEM and HYSCORE spectra are excellently reproduced by the DFTcalculated parameters of Ax x , A y y , and Az z characterizing the hyperfine interaction with the 11B nucleus and Cq and η characterizing the 11B quadrupolar interaction. Note that a direct measurement of the nuclear electric quadrupolar interaction via solid state NMR is not possible, as the 11B NMR signal of these spin-carrying molecules is broadened beyond detectability.5 Thus, the ESEEM technique is a convenient method for characterizing this interaction in borane/phosphane-based FLP-NO radicals. Finally, the absence of the nuclear Zeeman

TABLE IV. Nuclear electric hyperfine and quadrupolar coupling parameters and EFG principal axis orientations relative to A-tensor for the 11B nucleus in compounds 1, 2, and 3. Basis set for the DFT calculations—A-tensors: (1): B3LYP/TZ2P, (2): B3LYP/QZ4P, EFG-tensors: B97-D/def2-TZVP (mod.). Sample

Method

Axx (MHz)

Ayy (MHz)

Azz (MHz)

Cq (MHz)

η

α (deg)

β (deg)

γ (deg)

(1)

ESEEM DFTa(2)

... −11.0

... −10.9

... −6.21

... 1.30

... 0.57

... 40

... 7.5

... −17

(2)

ESEEM DFTb(1) DFTa(1)

−11.4 −14.8 −11.3

−11.2 −14.1 −11.2

−6.55 −9.7 −6.50

1.30 1.33 1.26

0.47 0.57 0.67

10 ... −4

175 ... 166

20 ... 55

(3)

ESEEM DFTb(1) DFTa(1)

−11.2 −11.1 −11.0

−11.0 −11.0 −10.9

−6.30 −6.27 −6.25

1.00 1.31 1.37

0.57 0.65 0.53

−32 −32 −19.4

178 164 109

64 64 86

a Based b Based

on gas phase optimized structures. on single crystal data.


124201-11


J. Chem. Phys. 142, 124201 (2015)

FIG. 11. Simulated HYSCORE spectra ((a), (b)) considering an axial electron-nucleus hyperfine interaction tensor for 11B, with A iso = −9.7 MHz and δ = 4.5 MHz, and isotropic hyperfine coupling tensors for 1H and 19F nuclei with A = 1.8 MHz and iso 0.8 MHz, respectively. (a) Neglecting 11B quadrupolar coupling interaction and (b) considering an anisotropic quadrupolar coupling interaction, with C q = 1.3 MHz and η = 0.57. (c) and (d) are experimental data from samples 2 and 3, respectively. The simulations and experimental measurements were performed for τ = 250 ns (see Figure 1 for τ definition). Dashed lines are guides to the eyes, representing the ν 1 = ν 2 line. The spectra are normalized by the maximum intensity.

frequencies of 10B, 11B, and 31P indicates that intermolecular dipolar interactions between the unpaired electrons and the nuclei of surrounding diamagnetic molecules of the host matrix can be neglected as well. Inspection of the crystal structure of the compound 1 indicates that closest intermolecular O · · · P and O · · · B distances are 6.9 and 6.8 Å, respectively. Evidently, these distances are too long for being detectable by pulsed EPR spectroscopy, at least in the presence of the (much stronger) intramolecular interactions. CONCLUSIONS

In summary, the complex nuclear and electronic spin interactions present in three new nitroxide radicals obtained by N,N addition of nitric oxide to vicinal borane-phosphane FLPs have been analyzed by continuous-wave and pulsed X-band EPR spectroscopy in the solid state. The two types of EPR experiments are found to be highly complementary. While the cw-EPR lineshapes are dominated by anisotropic g-tensor and nuclear-electron hyperfine interactions with the 14N and 31P nuclei, the electron spin-echo envelope modulation spectra are dominated by the hyperfine interactions with the 11B nuclei, allowing determination of the nuclear electric hyperfine coupling and quadrupolar interaction tensor values through simulations. For the simulation of both EPR lineshapes and ESEEM spectra, DFT-calculated values of the interaction parameters provide useful starting points for accomplishing “best-fit” agreement with the experimental data. The results obtained for three vicinal FLPs with different backbone structures are found to be quite similar, suggesting a common spectroscopic profile for this class of materials, which is, however, distinctly different from that of other nitroxide radicals.38–40 Overall, the combination of cw, pulsed EPR, and DFT-calculations provides a powerful spectroscopic approach for the spectroscopic characterization of these compounds.

ACKNOWLEDGMENTS

We acknowledge support by the Deutsche Forschungsgemeinschaft, Sonderforschungsbereich SFB 858, Synergetic Effects in Chemistry—From Additivity towards Reactivity. M.O. and H.E. acknowledge financial support by the Brazilian funding agencies São Paulo Research Foundation FAPESP (CEPID Project No. 2013/07793-6 and Grant No. 2013/23490-3) and CNPq (Universal Project No. 477053/2012-2). We thank the ETH Zürich for supporting the ADF calculations. 1Frustrated Lewis Pairs I, Uncovering and Understanding, Topics in Current

Chemistry Vol. 332, edited by D. Stephan and G. Erker (Springer, Heidelberg, New York, Dordrecht, London, 2013) and references therein. 2D. W. Stephan and G. Erker, Chem. Sci. 5, 2625 (2014), and references therein. 3(a) A. J. P. Cardenas, B. J. Culotta, T. H. Warren, S. Grimme, A. Stute, R. Fröhlich, G. Kehr, and G. Erker, Angew. Chem. 123, 7709 (2011); (b) M. Sajid, G. Kehr, T. Wiegand, H. Eckert, C. Schwickert, R. Pöttgen, A. J. P. Cardenas, T. H. Warren, R. Fröhlich, C. G. Daniliuc, and G. Erker, J. Am. Chem. Soc. 135, 8882 (2013). 4M. Sajid, A. Stute, A. J. P. Cardenas, B. J. Culotta, J. A. M. Hepperle, T. H. Warren, B. Schirmer, S. Grimme, A. Studer, C. G. Daniliuc, R. Fröhlich, J. L. Petersen, G. Kehr, and G. Erker, J. Am. Chem. Soc. 134, 10156 (2012). 5T. Wiegand, M. Sajid, G. Kehr, G. Erker, and H. Eckert, Solid State Nucl. Magn. Reson. 61-62, 19 (2014). 6A. Schweiger and G. Jeschke, Principles of Pulsed Electron Paramagnetic Resonance (University Press, Oxford, 2001). 7L. G. Rowan, E. L. Hahn, and W. B. Mims, Phys. Rev. A 138, 4 (1965); E. L. Hahn, Phys. Rev. 80, 580 (1950); W. B. Mims, Phys. Rev. B 5, 2409 (1972). 8P. Höfer, A. Grupp, G. Nebenfür, and M. Mehring, Chem. Phys. Lett. 132, 279 (1986). 9Time Domain Electron Spin Resonance, edited by L. Kevan and R. N. Schwartz (John Wiley & Sons, New York, 1979). 10Modern Pulsed and Continuous Wave Electron Spin Resonance, edited by L. Kevan and M. K. Bowman (John Wiley & Sons, 1990). 11S. A. Dikanov and Y. D. Tsvetkov, Electron Spin Echo Envelope Modulation (ESEEM) Spectroscopy (CRC Press, Inc., Boca Raton, FL, USA, 1992). 12High Resolution EPR, Biological Magnetic Resonance Vol. 28, edited by J. Harmer, G. Mitrikas, A. Schweiger, G. Hanson, and L. Berliner, (Springer Science+Business Media LLC, 2009) p. 13.


124201-12 13The


Theory of Magnetic Resonance, edited by C. P. Poole and H. A. Farach (Wiley-Interscience, New York, 1987). 14E. J. Reijerse and C. P. Keijzers, J. Magn. Reson. 71, 83 (1987). 15Y. Deligiannakis, L. Astrakas, G. Kordas, and R. A. Smith, Phys. Rev. B 58, 11420 (1998). 16A. Pöppl and L. Kevan, J. Phys. Chem. 100, 3387 (1996). 17H. Käss, J. Rautter, B. Bönigk, P. Höfer, and W. Lubitz, J. Phys. Chem. 99, 436 (1995). 18S. A. Dikanov and M. K. Bowman, J. Magn. Reson., Ser. A 116, 128 (1995). 19V. Kofman, O. Farver, I. Pecht, and D. Goldfarb, J. Am. Chem. Soc. 118, 1201 (1996). 20S. A. Dikanov, L. Xun, A. B. Karpiel, A. M. Tyryshkin, and M. K. Bowman, J. Am. Chem. Soc. 118, 8406 (1996). 21M. Gutjahr, R. Böttcher, and A. Pöppl, Appl. Magn. Reson. 22, 401 (2002). 22Y. Deligiannakis and A. W. Rutherford, Biochim. Biophys. Acta, Bioenerg. 1507, 226 (2001). 23S. Stoll and A. Schweiger, J. Magn. Reson. 178, 42 (2006). 24J. Tao, J. P. Perdew, V. N. Staroverov, and G. E. Scuseria, Phys. Rev. Lett. 91, 146401 (2003). 25S. Grimme, S. Ehrlich, and L. Goerigk, J. Comput. Chem. 32, 1456 (2011). 26F. Weigend and R. Ahlrichs, Phys. Chem. Chem. Phys. 7, 3297 (2005). 27R. Ahlrichs, F. Furche, and C. Hättig, TURBOMOLE, version 6.3 (Universität Karlsruhe, 2009); R. Ahlrichs, M. Bär, M. Häser, H. Horn, and C. Kölmel, Chem. Phys. Lett. 162, 165 (1989). 28O. Treutler and R. Ahlrichs, J. Chem. Phys. 102, 346 (1995). 29K. Eichkorn, O. Treutler, H. Öhm, M. Häser, and R. Ahlrichs, Chem. Phys. Lett. 240, 283 (1995); K. Eichkorn, F. Weigend, O. Treutler, and R. Ahlrichs, Theor. Chem. Acc. 97, 119 (1997).

J. Chem. Phys. 142, 124201 (2015) 30ADF2013,

SCM, Theoretical Chemistry (Vrije Universiteit, Amsterdam, The Netherlands, 2013), http://www.scm.com; G. te Velde, F. M. Bickelhaupt, E. J. Baerends, C. F. Guerra, S. J. A. van Gisbergen, J. G. Snijders, and T. Ziegler, J. Comput. Chem. 22, 931 (2001); C. F. Guerra, J. G. Snijders, G. te Velde, and E. J. Baerends, Theor. Chem. Acc. 99, 391 (1998). 31A. D. Becke, J. Chem. Phys. 98, 5648 (1993); P. J. Stephens, F. J. Devlin, C. F. Chabalowski, and M. J. Frisch, J. Phys. Chem. 98, 11623 (1994). 32E. Van Lenthe and E. J. Baerends, J. Comput. Chem. 24, 1142 (2003). 33W. T. Pennington, J. Appl. Crystallogr. 32, 1028 (1999). 34E. van Lenthe, A. Ehlers, and E.-J. Baerends, J. Chem. Phys. 110, 8943 (1999); E. van Lenthe, J. G. Snijders, and E. J. Baerends, J. Chem. Phys. 105, 6505 (1996); E. van Lenthe, E. J. Baerends, and J. G. Snijders, J. Chem. Phys. 101, 9783 (1994); E. van Lenthe, J. Baerends, and J. G. Snijders, J. Chem. Phys. 99, 4597 (1993). 35M. J. Frisch et al.,  09, Gaussian, Inc., Wallingford, CT, 2009. 36T. Wiegand, H. Eckert, O. Ekkert, R. Fröhlich, G. Kehr, G. Erker, and S. Grimme, J. Am. Chem. Soc. 134, 4236 (2012). 37S. Adiga, D. Aebi, and D. L. Bryce, Can. J. Chem. 85, 496 (2007). 38D. E. Budil, K. A. Earle, and J. H. Freed, J. Phys. Chem. 97, 1294 (1993). 39B. Dizikovski, D. Tipikin, L. Vsevolod, K. Earle, and J. Freed, Phys. Chem. Chem. Phys. 11, 6676 (2009). 40M. Tabak, A. Alonso, and O. R. Nascimento, J. Chem. Phys. 79, 1176 (1983). 41S. Stoll, C. Calle, G. Mitrikas, and A. Schweiger, J. Magn. Reson. 177, 93 (2005). 42See supplementary material at http://dx.doi.org/10.1063/1.4916066 for Test simulations of cw-EPR spectra on TEMPO grafted on silica; echo detected EPR spectra, and DFT calculation results using different electron correlation functionals.


Frustrated Lewis pairs: a metal-free landmark.

Frustrated Lewis pairs: from concept to catalysis.

Frustrated N-heterocyclic carbene-silylium ion Lewis pairs.

Enabling catalytic ketone hydrogenation by frustrated Lewis pairs.

Metal-free electrocatalytic hydrogen oxidation using frustrated Lewis pairs and carbon-based Lewis acids.

P frustrated Lewis pairs by the unique 1,1-carbozirconation reactions.

Stoichiometric and catalytic inter- and intramolecular hydroamination of terminal alkynes by frustrated Lewis pairs.

Probing the association of frustrated phosphine-borane Lewis pairs in solution by NMR spectroscopy.

Catalytic B-N Dehydrogenation Using Frustrated Lewis Pairs: Evidence for a Chain-Growth Coupling Mechanism.

N frustrated Lewis pairs and the hydrogenation of carbon dioxide.

A Strategy to Control the Reactivation of Frustrated Lewis Pairs from Shelf-Stable Carbene Borane Complexes.

P Frustrated Lewis Pairs: Refinement of the Spin-Hamiltonian Parameters by Field- and Temperature-Dependent Pulsed EPR Spectroscopy.

An electrochemical study of frustrated Lewis pairs: a metal-free route to hydrogen oxidation.

N-methylacridinium salts: carbon Lewis acids in frustrated Lewis pairs for σ-bond activation and catalytic reductions.

A highly enantioselective hydrogenation of silyl enol ethers catalyzed by chiral frustrated Lewis pairs.

borane Lewis pairs by a polymerization study.

borane Lewis pairs.

Frustrated Lewis pairs as molecular receptors: colorimetric and electrochemical detection of nitrous oxide.

Frustrated Lewis pair chemistry.

Are intramolecular frustrated Lewis pairs also intramolecular catalysts? A theoretical study on H2 activation.

A computational and conceptual DFT study on the mechanism of hydrogen activation by novel frustrated Lewis pairs.

Reducing CO₂ to methanol using frustrated Lewis pairs: on the mechanism of phosphine-borane-mediated hydroboration of CO₂.

Formylborane formation with frustrated Lewis pair templates.

Facile Protocol for Water-Tolerant "Frustrated Lewis Pair"-Catalyzed Hydrogenation.